Temporal Patterns in Multi-modal Social Interaction between Elderly Users and Service Robot

Social interaction, especially for older people living alone is a challenge currently facing human-robot interaction (HRI). User interfaces to manage service robots in home environments need to be tailored for older people. Multi-modal interfaces providing users with more than one communication option seem promising. There has been little research on user preference towards HRI interfaces; most studies have focused on utility and functionality of the interface. In this paper, we took both objective observations and participants’ opinions into account in studying older users with a robot partner. Our study was under the framework of the EU FP7 Robot-Era Project. The developed dual-modal robot interface offered older users options of speech or touch screen to perform tasks. Fifteen people aged from 70 to 89 years old, participated. We analyzed the spontaneous actions of the participants, including their attentional activities (eye contacts) and conversational activities, the temporal characteristics (timestamps, duration of events, event transitions) of these social behaviours, as well as questionnaires. This combination of data distinguishes it from other studies that focused on questionnaire ratings only. There were three main findings. First, the design of the Robot-Era interface was very acceptable for older users. Secondly, most older people used both speech and tablet to perform the food delivery service, with no difference in their preferences towards either. Thirdly, these older people had frequent and long-duration eye contact with the robot during their conversations, showing patience when expecting the robot to respond. They enjoyed the service. Overall, social engagement with the robot demonstrated by older people was no different from what might be expected towards a human partner. This study is an early attempt to reveal the social connections between human beings and a personal robot in real life. Our observations and findings should inspire new insights in HRI research and eventually contribute to next-generation intelligent robot development